Every application platform now drafts applications from a description. The useful question is where the checks sit.
In May 2025 a security researcher scanned 1,645 applications published with a consumer AI app builder and found that 170 of them, 10.3 percent, exposed data through missing or insufficient row-level security; the finding was recorded as CVE-2025-48757 [1]. The applications worked; the access rules were absent. The same year Gartner's definition of an enterprise low-code platform listed AI-assisted development beside data virtualisation, governance and security among the capabilities a platform must have [2], and Gartner predicted that over 40 percent of agentic AI projects would be cancelled by the end of 2027 for reasons that include inadequate risk controls [3].
When an external model drafts, the description and the structure cross: names of screens, components, levels and roles; handler definitions; criteria as text. Every row from your systems stays inside the platform, and the exchange is attested and logged. The choice of model is yours; the gates are the same for every model and for drafts the platform makes alone.
The first session builds the first screen on your live rows.